http://d2l.ai/chapter_convolutional-neural-networks/pooling.html

On the third paragraph, if Z[i, j] = X[i, j + 1] then isnâ€™t Z the result of shifting X one pixel to the **left**? Z[i, 0] = X[i, 1] so the second column of X becomes the first column of Z.

the textbook is correct in fact

Is there a possibility that for 6.5.5 1), you can actually create a different convolutional layer which gives the same results as an average pooling? If so, how can we do this?

Looking for a hint about 6.5.5. exercise 6. Can anyone provide relevant source to such pooling technique between average and maximum pooling?